src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @file src/Content/Text/HTML.php
   4  */
   5
   6 namespace Friendica\Content\Text;
   7
   8 use DOMDocument;
   9 use DOMXPath;
  10 use Friendica\Content\Feature;
  11 use Friendica\Core\Addon;
  12 use Friendica\Core\L10n;
  13 use Friendica\Core\Config;
  14 use Friendica\Core\PConfig;
  15 use Friendica\Core\Protocol;
  16 use Friendica\Core\Renderer;
  17 use Friendica\Database\DBA;
  18 use Friendica\Model\Contact;
  19 use Friendica\Util\Network;
  20 use Friendica\Util\Proxy as ProxyUtils;
  21 use Friendica\Util\Strings;
  22 use Friendica\Util\XML;
  23 use League\HTMLToMarkdown\HtmlConverter;
  24
  25 class HTML
  26 {
  27         public static function sanitizeCSS($input)
  28         {
  29                 $cleaned = "";
  30
  31                 $input = strtolower($input);
  32
  33                 for ($i = 0; $i < strlen($input); $i++) {
  34                         $char = substr($input, $i, 1);
  35
  36                         if (($char >= "a") && ($char <= "z")) {
  37                                 $cleaned .= $char;
  38                         }
  39
  40                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  41                                 $cleaned .= $char;
  42                         }
  43                 }
  44
  45                 return $cleaned;
  46         }
  47
  48         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  49         {
  50                 do {
  51                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  52                 } while ($done);
  53         }
  54
  55         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  56         {
  57                 $savestart = str_replace('$', '\x01', $startbb);
  58                 $replace = false;
  59
  60                 $xpath = new DOMXPath($doc);
  61
  62                 $list = $xpath->query("//" . $tag);
  63                 foreach ($list as $node) {
  64                         $attr = [];
  65                         if ($node->attributes->length) {
  66                                 foreach ($node->attributes as $attribute) {
  67                                         $attr[$attribute->name] = $attribute->value;
  68                                 }
  69                         }
  70
  71                         $replace = true;
  72
  73                         $startbb = $savestart;
  74
  75                         $i = 0;
  76
  77                         foreach ($attributes as $attribute => $value) {
  78                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  79                                 if (strpos('*' . $startbb, '$1') > 0) {
  80                                         if ($replace && (@$attr[$attribute] != '')) {
  81                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  82
  83                                                 // If nothing could be changed
  84                                                 if ($count == 0) {
  85                                                         $replace = false;
  86                                                 }
  87                                         } else {
  88                                                 $replace = false;
  89                                         }
  90                                 } else {
  91                                         if (@$attr[$attribute] != $value) {
  92                                                 $replace = false;
  93                                         }
  94                                 }
  95                         }
  96
  97                         if ($replace) {
  98                                 $StartCode = $doc->createTextNode($startbb);
  99                                 $EndCode = $doc->createTextNode($endbb);
 100
 101                                 $node->parentNode->insertBefore($StartCode, $node);
 102
 103                                 if ($node->hasChildNodes()) {
 104                                         foreach ($node->childNodes as $child) {
 105                                                 $newNode = $child->cloneNode(true);
 106                                                 $node->parentNode->insertBefore($newNode, $node);
 107                                         }
 108                                 }
 109
 110                                 $node->parentNode->insertBefore($EndCode, $node);
 111                                 $node->parentNode->removeChild($node);
 112                         }
 113                 }
 114
 115                 return $replace;
 116         }
 117
 118         /**
 119          * Made by: ike@piratenpartei.de
 120          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 121          *                                      https://github.com/annando/Syncom
 122          *
 123          * @brief Converter for HTML to BBCode
 124          * @param string $message
 125          * @param string $basepath
 126          * @return string
 127          */
 128         public static function toBBCode($message, $basepath = '')
 129         {
 130                 $message = str_replace("\r", "", $message);
 131
 132                 // Removing code blocks before the whitespace removal processing below
 133                 $codeblocks = [];
 134                 $message = preg_replace_callback(
 135                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 136                         function ($matches) use (&$codeblocks) {
 137                                 $return = '[codeblock-' . count($codeblocks) . ']';
 138
 139                                 $prefix = '[code]';
 140                                 if ($matches[1] != '') {
 141                                         $prefix = '[code=' . $matches[1] . ']';
 142                                 }
 143
 144                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 145                                 return $return;
 146                         },
 147                         $message
 148                 );
 149
 150                 $message = str_replace(
 151                         [
 152                                 "<li><p>",
 153                                 "</p></li>",
 154                         ],
 155                         [
 156                                 "<li>",
 157                                 "</li>",
 158                         ],
 159                         $message
 160                 );
 161
 162                 // remove namespaces
 163                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 164                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 165
 166                 $doc = new DOMDocument();
 167                 $doc->preserveWhiteSpace = false;
 168
 169                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 170
 171                 @$doc->loadHTML($message);
 172
 173                 XML::deleteNode($doc, 'style');
 174                 XML::deleteNode($doc, 'head');
 175                 XML::deleteNode($doc, 'title');
 176                 XML::deleteNode($doc, 'meta');
 177                 XML::deleteNode($doc, 'xml');
 178                 XML::deleteNode($doc, 'removeme');
 179
 180                 $xpath = new DomXPath($doc);
 181                 $list = $xpath->query("//pre");
 182                 foreach ($list as $node) {
 183                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 184                 }
 185
 186                 $message = $doc->saveHTML();
 187                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 188                 $message = preg_replace('= [\s]*=i', " ", $message);
 189                 @$doc->loadHTML($message);
 190
 191                 self::tagToBBCode($doc, 'html', [], "", "");
 192                 self::tagToBBCode($doc, 'body', [], "", "");
 193
 194                 // Outlook-Quote - Variant 1
 195                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 196
 197                 // Outlook-Quote - Variant 2
 198                 self::tagToBBCode(
 199                         $doc,
 200                         'div',
 201                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 202                         '[quote]',
 203                         '[/quote]'
 204                 );
 205
 206                 // MyBB-Stuff
 207                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 208                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 209                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 210
 211                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 212                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 213                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 214                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 215                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 216                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 217                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 218                  */
 219                 // Untested
 220                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 221                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 222                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 223
 224                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 225
 226                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 227                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 228                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 229                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 230                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 231                 // Test
 232                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 233                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 234                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 235
 236                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 237                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 238                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 239                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 240                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 241                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 242                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 243                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 244
 245                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 246                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 247
 248                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 249
 250                 self::tagToBBCode($doc, 'br', [], "\n", '');
 251
 252                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 253                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 254
 255                 self::tagToBBCode($doc, 'span', [], "", "");
 256
 257                 self::tagToBBCode($doc, 'span', [], "", "");
 258                 self::tagToBBCode($doc, 'pre', [], "", "");
 259
 260                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 261                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 262
 263                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 264                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 265                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 266
 267                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 268
 269                 self::tagToBBCode($doc, 'table', [], "", "");
 270                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 271                 self::tagToBBCode($doc, 'td', [], "\t", "");
 272                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 273                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 274                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 275                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 276                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 277                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 282
 283                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 284                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 285                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 286                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 287                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 288                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 289
 290                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 291                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 292
 293                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 294                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 295
 296
 297                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 298                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 299                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 300
 301                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 302                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 303
 304                 $message = $doc->saveHTML();
 305
 306                 // I'm removing something really disturbing
 307                 // Don't know exactly what it is
 308                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 309
 310                 $message = str_replace("&nbsp;", " ", $message);
 311
 312                 // removing multiple DIVs
 313                 $message = preg_replace('=\r *\r=i', "\n", $message);
 314                 $message = str_replace("\r", "\n", $message);
 315
 316                 Addon::callHooks('html2bbcode', $message);
 317
 318                 $message = strip_tags($message);
 319
 320                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 321
 322                 $message = str_replace(["<"], ["&lt;"], $message);
 323
 324                 // remove quotes if they don't make sense
 325                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 326
 327                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 328                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 329
 330                 do {
 331                         $oldmessage = $message;
 332                         $message = str_replace("\n \n", "\n\n", $message);
 333                 } while ($oldmessage != $message);
 334
 335                 do {
 336                         $oldmessage = $message;
 337                         $message = str_replace("\n\n\n", "\n\n", $message);
 338                 } while ($oldmessage != $message);
 339
 340                 do {
 341                         $oldmessage = $message;
 342                         $message = str_replace(
 343                                 [
 344                                 "[/size]\n\n",
 345                                 "\n[hr]",
 346                                 "[hr]\n",
 347                                 "\n[list",
 348                                 "[/list]\n",
 349                                 "\n[/",
 350                                 "[list]\n",
 351                                 "[list=1]\n",
 352                                 "\n[*]"],
 353                                 [
 354                                 "[/size]\n",
 355                                 "[hr]",
 356                                 "[hr]",
 357                                 "[list",
 358                                 "[/list]",
 359                                 "[/",
 360                                 "[list]",
 361                                 "[list=1]",
 362                                 "[*]"],
 363                                 $message
 364                         );
 365                 } while ($message != $oldmessage);
 366
 367                 $message = str_replace(
 368                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 369                         ['[b]', '[/b]', '[i]', '[/i]'],
 370                         $message
 371                 );
 372
 373                 // Handling Yahoo style of mails
 374                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 375
 376                 // Restore code blocks
 377                 $message = preg_replace_callback(
 378                         '#\[codeblock-([0-9]+)\]#iU',
 379                         function ($matches) use ($codeblocks) {
 380                                 $return = '';
 381                                 if (isset($codeblocks[intval($matches[1])])) {
 382                                         $return = $codeblocks[$matches[1]];
 383                                 }
 384                                 return $return;
 385                         },
 386                         $message
 387                 );
 388
 389                 $message = trim($message);
 390
 391                 if ($basepath != '') {
 392                         $message = self::qualifyURLs($message, $basepath);
 393                 }
 394
 395                 return $message;
 396         }
 397
 398         /**
 399          * @brief Sub function to complete incomplete URL
 400          *
 401          * @param array  $matches  Result of preg_replace_callback
 402          * @param string $basepath Basepath that is used to complete the URL
 403          *
 404          * @return string The expanded URL
 405          */
 406         private static function qualifyURLsSub($matches, $basepath)
 407         {
 408                 $base = parse_url($basepath);
 409                 unset($base['query']);
 410                 unset($base['fragment']);
 411
 412                 $link = $matches[0];
 413                 $url = $matches[1];
 414
 415                 $parts = array_merge($base, parse_url($url));
 416                 $url2 = Network::unparseURL($parts);
 417
 418                 return str_replace($url, $url2, $link);
 419         }
 420
 421         /**
 422          * @brief Complete incomplete URLs in BBCode
 423          *
 424          * @param string $body     Body with URLs
 425          * @param string $basepath Base path that is used to complete the URL
 426          *
 427          * @return string Body with expanded URLs
 428          */
 429         private static function qualifyURLs($body, $basepath)
 430         {
 431                 $URLSearchString = "^\[\]";
 432
 433                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 434                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 435                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 436                         "/\[img\](.*?)\[\/img\]/ism",
 437                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 438                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 439                         "/\[video\](.*?)\[\/video\]/ism",
 440                         "/\[audio\](.*?)\[\/audio\]/ism",
 441                 ];
 442
 443                 foreach ($matches as $match) {
 444                         $body = preg_replace_callback(
 445                                 $match,
 446                                 function ($match) use ($basepath) {
 447                                         return self::qualifyURLsSub($match, $basepath);
 448                                 },
 449                                 $body
 450                         );
 451                 }
 452                 return $body;
 453         }
 454
 455         private static function breakLines($line, $level, $wraplength = 75)
 456         {
 457                 if ($wraplength == 0) {
 458                         $wraplength = 2000000;
 459                 }
 460
 461                 $wraplen = $wraplength - $level;
 462
 463                 $newlines = [];
 464
 465                 do {
 466                         $oldline = $line;
 467
 468                         $subline = substr($line, 0, $wraplen);
 469
 470                         $pos = strrpos($subline, ' ');
 471
 472                         if ($pos == 0) {
 473                                 $pos = strpos($line, ' ');
 474                         }
 475
 476                         if (($pos > 0) && strlen($line) > $wraplen) {
 477                                 $newline = trim(substr($line, 0, $pos));
 478                                 if ($level > 0) {
 479                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 480                                 }
 481
 482                                 $newlines[] = $newline . " ";
 483                                 $line = substr($line, $pos + 1);
 484                         }
 485                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 486
 487                 if ($level > 0) {
 488                         $line = str_repeat(">", $level) . ' ' . $line;
 489                 }
 490
 491                 $newlines[] = $line;
 492
 493                 return implode($newlines, "\n");
 494         }
 495
 496         private static function quoteLevel($message, $wraplength = 75)
 497         {
 498                 $lines = explode("\n", $message);
 499
 500                 $newlines = [];
 501                 $level = 0;
 502                 foreach ($lines as $line) {
 503                         $line = trim($line);
 504                         $startquote = false;
 505                         while (strpos("*" . $line, '[quote]') > 0) {
 506                                 $level++;
 507                                 $pos = strpos($line, '[quote]');
 508                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 509                                 $startquote = true;
 510                         }
 511
 512                         $currlevel = $level;
 513
 514                         while (strpos("*" . $line, '[/quote]') > 0) {
 515                                 $level--;
 516                                 if ($level < 0) {
 517                                         $level = 0;
 518                                 }
 519
 520                                 $pos = strpos($line, '[/quote]');
 521                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 522                         }
 523
 524                         if (!$startquote || ($line != '')) {
 525                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 526                         }
 527                 }
 528
 529                 return implode($newlines, "\n");
 530         }
 531
 532         private static function collectURLs($message)
 533         {
 534                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 535                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 536
 537                 $urls = [];
 538                 foreach ($result as $treffer) {
 539                         $ignore = false;
 540
 541                         // A list of some links that should be ignored
 542                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 543                                 "//plus.google.com/", "//twitter.com/"];
 544                         foreach ($list as $listitem) {
 545                                 if (strpos($treffer[1], $listitem) !== false) {
 546                                         $ignore = true;
 547                                 }
 548                         }
 549
 550                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 551                                 $ignore = false;
 552                         }
 553
 554                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 555                                 $ignore = false;
 556                         }
 557
 558                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 559                                 $ignore = false;
 560                         }
 561
 562                         if (!$ignore) {
 563                                 $urls[$treffer[1]] = $treffer[1];
 564                         }
 565                 }
 566
 567                 return $urls;
 568         }
 569
 570         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 571         {
 572                 $message = str_replace("\r", "", $html);
 573
 574                 $doc = new DOMDocument();
 575                 $doc->preserveWhiteSpace = false;
 576
 577                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 578
 579                 @$doc->loadHTML($message);
 580
 581                 $xpath = new DOMXPath($doc);
 582                 $list = $xpath->query("//pre");
 583                 foreach ($list as $node) {
 584                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 585                 }
 586
 587                 $message = $doc->saveHTML();
 588                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 589                 $message = preg_replace('= [\s]*=i', " ", $message);
 590
 591                 // Collecting all links
 592                 $urls = self::collectURLs($message);
 593
 594                 @$doc->loadHTML($message);
 595
 596                 self::tagToBBCode($doc, 'html', [], '', '');
 597                 self::tagToBBCode($doc, 'body', [], '', '');
 598
 599                 // MyBB-Auszeichnungen
 600                 /*
 601                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 602                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 603                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 604
 605                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 606                   self::node2BBCode($doc, 'b', array(), '*', '*');
 607                   self::node2BBCode($doc, 'i', array(), '/', '/');
 608                   self::node2BBCode($doc, 'u', array(), '_', '_');
 609                  */
 610
 611                 if ($compact) {
 612                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 613                 } else {
 614                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 615                 }
 616
 617                 self::tagToBBCode($doc, 'br', [], "\n", '');
 618
 619                 self::tagToBBCode($doc, 'span', [], "", "");
 620                 self::tagToBBCode($doc, 'pre', [], "", "");
 621                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 622                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 623
 624                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 625                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 626                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 627
 628                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 629
 630                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 631                 self::tagToBBCode($doc, 'td', [], "\t", "");
 632
 633                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 639
 640                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 641                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 642                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 643                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 644                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 645                 //self::node2BBCode($doc, 'img', array(), '', '');
 646                 if (!$compact) {
 647                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 648                 } else {
 649                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 650                 }
 651
 652                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 653
 654                 $message = $doc->saveHTML();
 655
 656                 if (!$compact) {
 657                         $message = str_replace("[img]", "", $message);
 658                         $message = str_replace("[/img]", "", $message);
 659                 }
 660
 661                 // was ersetze ich da?
 662                 // Irgendein stoerrisches UTF-Zeug
 663                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 664
 665                 $message = str_replace("&nbsp;", " ", $message);
 666
 667                 // Aufeinanderfolgende DIVs
 668                 $message = preg_replace('=\r *\r=i', "\n", $message);
 669                 $message = str_replace("\r", "\n", $message);
 670
 671                 $message = strip_tags($message);
 672
 673                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 674
 675                 if (!$compact && ($message != '')) {
 676                         foreach ($urls as $id => $url) {
 677                                 if ($url != '' && strpos($message, $url) === false) {
 678                                         $message .= "\n" . $url . ' ';
 679                                 }
 680                         }
 681                 }
 682
 683                 $message = str_replace("\n«", "«\n", $message);
 684                 $message = str_replace("»\n", "\n»", $message);
 685
 686                 do {
 687                         $oldmessage = $message;
 688                         $message = str_replace("\n\n\n", "\n\n", $message);
 689                 } while ($oldmessage != $message);
 690
 691                 $message = self::quoteLevel(trim($message), $wraplength);
 692
 693                 return trim($message);
 694         }
 695
 696         /**
 697          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 698          * compatibility with Diaspora in spite of the Markdown standards.
 699          *
 700          * @param string $html
 701          * @return string
 702          */
 703         public static function toMarkdown($html)
 704         {
 705                 $converter = new HtmlConverter(['hard_break' => true]);
 706                 $markdown = $converter->convert($html);
 707
 708                 return $markdown;
 709         }
 710
 711         /**
 712          * @brief Convert video HTML to BBCode tags
 713          *
 714          * @param string $s
 715          */
 716         public static function toBBCodeVideo($s)
 717         {
 718                 $s = preg_replace(
 719                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 720                         '[youtube]$2[/youtube]',
 721                         $s
 722                 );
 723
 724                 $s = preg_replace(
 725                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 726                         '[youtube]$2[/youtube]',
 727                         $s
 728                 );
 729
 730                 $s = preg_replace(
 731                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 732                         '[vimeo]$2[/vimeo]',
 733                         $s
 734                 );
 735
 736                 return $s;
 737         }
 738
 739         /**
 740          * transform link href and img src from relative to absolute
 741          *
 742          * @param string $text
 743          * @param string $base base url
 744          * @return string
 745          */
 746         public static function relToAbs($text, $base)
 747         {
 748                 if (empty($base)) {
 749                         return $text;
 750                 }
 751
 752                 $base = rtrim($base, '/');
 753
 754                 $base2 = $base . "/";
 755
 756                 // Replace links
 757                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 758                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 759                 $text = preg_replace($pattern, $replace, $text);
 760
 761                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 762                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 763                 $text = preg_replace($pattern, $replace, $text);
 764
 765                 // Replace images
 766                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 767                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 768                 $text = preg_replace($pattern, $replace, $text);
 769
 770                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 771                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 772                 $text = preg_replace($pattern, $replace, $text);
 773
 774
 775                 // Done
 776                 return $text;
 777         }
 778
 779         /**
 780          * return div element with class 'clear'
 781          * @return string
 782          * @deprecated
 783          */
 784         public static function clearDiv()
 785         {
 786                 return '<div class="clear"></div>';
 787         }
 788
 789         /**
 790          * Loader for infinite scrolling
 791          * @return string html for loader
 792          */
 793         public static function scrollLoader()
 794         {
 795                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 796                 return Renderer::replaceMacros($tpl, [
 797                         'wait' => L10n::t('Loading more entries...'),
 798                         'end' => L10n::t('The end')
 799                 ]);
 800         }
 801
 802         /**
 803          * Get html for contact block.
 804          *
 805          * @template contact_block.tpl
 806          * @hook contact_block_end (contacts=>array, output=>string)
 807          * @return string
 808          */
 809         public static function contactBlock()
 810         {
 811                 $o = '';
 812                 $a = get_app();
 813
 814                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 815                 if ($shown == 0) {
 816                         return;
 817                 }
 818
 819                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 820                         return $o;
 821                 }
 822
 823                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 824                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 825                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 826                                         AND `network` IN ('%s', '%s', '%s')",
 827                         intval($a->profile['uid']),
 828                         DBA::escape(Protocol::DFRN),
 829                         DBA::escape(Protocol::OSTATUS),
 830                         DBA::escape(Protocol::DIASPORA)
 831                 );
 832
 833                 if (DBA::isResult($r)) {
 834                         $total = intval($r[0]['total']);
 835                 }
 836
 837                 if (!$total) {
 838                         $contacts = L10n::t('No contacts');
 839                         $micropro = null;
 840                 } else {
 841                         // Splitting the query in two parts makes it much faster
 842                         $r = q("SELECT `id` FROM `contact`
 843                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 844                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 845                                                 AND `network` IN ('%s', '%s', '%s')
 846                                         ORDER BY RAND() LIMIT %d",
 847                                 intval($a->profile['uid']),
 848                                 DBA::escape(Protocol::DFRN),
 849                                 DBA::escape(Protocol::OSTATUS),
 850                                 DBA::escape(Protocol::DIASPORA),
 851                                 intval($shown)
 852                         );
 853
 854                         if (DBA::isResult($r)) {
 855                                 $contacts = [];
 856                                 foreach ($r as $contact) {
 857                                         $contacts[] = $contact["id"];
 858                                 }
 859
 860                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 861                                         DBA::escape(implode(",", $contacts))
 862                                 );
 863
 864                                 if (DBA::isResult($r)) {
 865                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 866                                         $micropro = [];
 867                                         foreach ($r as $rr) {
 868                                                 $micropro[] = self::micropro($rr, true, 'mpfriend');
 869                                         }
 870                                 }
 871                         }
 872                 }
 873
 874                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 875                 $o = Renderer::replaceMacros($tpl, [
 876                         '$contacts' => $contacts,
 877                         '$nickname' => $a->profile['nickname'],
 878                         '$viewcontacts' => L10n::t('View Contacts'),
 879                         '$micropro' => $micropro,
 880                 ]);
 881
 882                 $arr = ['contacts' => $r, 'output' => $o];
 883
 884                 Addon::callHooks('contact_block_end', $arr);
 885
 886                 return $o;
 887         }
 888
 889         /**
 890          * @brief Format contacts as picture links or as texxt links
 891          *
 892          * @param array $contact Array with contacts which contains an array with
 893          *      int 'id' => The ID of the contact
 894         *       int 'uid' => The user ID of the user who owns this data
 895         *       string 'name' => The name of the contact
 896         *       string 'url' => The url to the profile page of the contact
 897         *       string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 898         *       string 'network' => The network to which the contact belongs to
 899         *       string 'thumb' => The contact picture
 900         *       string 'click' => js code which is performed when clicking on the contact
 901         * @param boolean $redirect If true try to use the redir url if it's possible
 902         * @param string $class CSS class for the
 903         * @param boolean $textmode If true display the contacts as text links
 904         *       if false display the contacts as picture links
 905
 906         * @return string Formatted html
 907         */
 908         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 909         {
 910                 // Use the contact URL if no address is available
 911                 if (!x($contact, "addr")) {
 912                         $contact["addr"] = $contact["url"];
 913                 }
 914
 915                 $url = $contact['url'];
 916                 $sparkle = '';
 917                 $redir = false;
 918
 919                 if ($redirect) {
 920                         $url = Contact::magicLink($contact['url']);
 921                         if (strpos($url, 'redir/') === 0) {
 922                                 $sparkle = ' sparkle';
 923                         }
 924                 }
 925
 926                 // If there is some js available we don't need the url
 927                 if (x($contact, 'click')) {
 928                         $url = '';
 929                 }
 930
 931                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 932                         '$click' => defaults($contact, 'click', ''),
 933                         '$class' => $class,
 934                         '$url' => $url,
 935                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 936                         '$name' => $contact['name'],
 937                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 938                         '$parkle' => $sparkle,
 939                         '$redir' => $redir
 940                 ]);
 941         }
 942
 943         /**
 944          * Search box.
 945          *
 946          * @param string $s     Search query.
 947          * @param string $id    HTML id
 948          * @param string $url   Search url.
 949          * @param bool   $save  Show save search button.
 950          * @param bool   $aside Display the search widgit aside.
 951          *
 952          * @return string Formatted HTML.
 953          */
 954         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 955         {
 956                 $mode = 'text';
 957
 958                 if (strpos($s, '#') === 0) {
 959                         $mode = 'tag';
 960                 }
 961                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 962
 963                 $values = [
 964                                 '$s' => htmlspecialchars($s),
 965                                 '$id' => $id,
 966                                 '$action_url' => $url,
 967                                 '$search_label' => L10n::t('Search'),
 968                                 '$save_label' => $save_label,
 969                                 '$savedsearch' => 'savedsearch',
 970                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 971                                 '$mode' => $mode
 972                         ];
 973
 974                 if (!$aside) {
 975                         $values['$searchoption'] = [
 976                                                 L10n::t("Full Text"),
 977                                                 L10n::t("Tags"),
 978                                                 L10n::t("Contacts")];
 979
 980                         if (Config::get('system', 'poco_local_search')) {
 981                                 $values['$searchoption'][] = L10n::t("Forums");
 982                         }
 983                 }
 984
 985                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 986         }
 987
 988         /**
 989          * Replace naked text hyperlink with HTML formatted hyperlink
 990          *
 991          * @param string $s
 992          */
 993         public static function toLink($s)
 994         {
 995                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 996                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 997                 return $s;
 998         }
 999
1000         /**
1001          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
1002          *
1003          * Reasons are expected to have been translated already.
1004          *
1005          * @param string $html
1006          * @param array  $reasons
1007          * @return string
1008          */
1009         public static function applyContentFilter($html, array $reasons)
1010         {
1011                 if (count($reasons)) {
1012                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1013                         $html = Renderer::replaceMacros($tpl, [
1014                                 '$reasons'   => $reasons,
1015                                 '$rnd'       => Strings::getRandomHex(8),
1016                                 '$openclose' => L10n::t('Click to open/close'),
1017                                 '$html'      => $html
1018                         ]);
1019                 }
1020
1021                 return $html;
1022         }
1023
1024         /**
1025          * replace html amp entity with amp char
1026          * @param string $s
1027          * @return string
1028          */
1029         public static function unamp($s)
1030         {
1031                 return str_replace('&amp;', '&', $s);
1032         }
1033 }